摘要

Iterative power difference balancing (IPDB) has recently been proposed as a first real-time dynamic spectrum management (RT-DSM) algorithm. It consists of a primal coordinate ascent approach where each coordinate step is performed using an exhaustive discrete grid line search. In this paper we present an iterative convex approximation based approach to perform the coordinate ascent search so as to reduce its computational complexity. By exploiting the problem structure, a closed-form solution is derived for the convex approximations. The resulting RT-DSM algorithm is referred to as fast IPDB (F-IPDB). Compared to IPDB, F-IPDB exhibits similar data rate performance with significantly reduced computational complexity, while also providing smoother final transmit spectra.

  • 出版日期2014-5